The reason you haven't seen any clear answers is that there aren't any; the variables involved with the history of the polyswitch fuse, the draw from the USB, how often the Pi is switched on and how long it rests between, what USB devices are connected and when, are all interrelated in a highly complex fashion.
is everything you need to know about how polyswitch devices work.
Tyco Electronics wrote:... we can clearly see that even after a number of hours the device resistance is still greater than the initial resistance. Over an extended period of time, the resistance will continue to fall and will eventually approach the initial resistance.However, since this time can be days, months, or years, it is not practical to expect that the device resistance will reach the original value for operational purposes. Therefore, when PolySwitch devices are being developed, this "trip jump" or "reflow jump" is taken into consideration when determining the hold current. This increase in resistance is defined as R1MAX and is measured one hour after the thermal event. It should be noted that these trip jumps are non-cumulative over sequential trip events.
Which means that even many trips over an extended period should not cause a problem. The polyswitch should be back in spec within 1 hour. My feeling is that for whatever reason, and it might be just that the input voltage is below 5V, the non-trip increase in resistance (see figure 3) is sufficient to cause your USB devices to fail. You should measure the voltage either side of the polyswitch with respect to TP2. If the higher of the two voltages are below 4.75V, or rather close to it, investigate a new USB cable or PSU. If the lower voltage is below 4.75V then it may cause problems with the USB devices, especially if you are using a PS/2 keyboard with a USB adapter. However USB devices should
work with voltages down to 4.4V. The difference between the values should be no more than 0.2V, or maybe 0.3V if you are working the Pi very hard and using thirsty USB devices; if it is more then you have a faulty unit. (See this TE document
. I believe F3 is a microSMD-075 or a close equivalent, with a maximum (R1MAX) resistance of 0.4 ohms.)
It may be instructive to invest in some freezer spray and cool just F3 at 30 second intervals. If that stops your USB devices dying it means F3 is the cause of the problem. It doesn't mean that it is faulty though; the problem could be that something else is out of spec, making F3 more critical.